The relationship of mental illness and violence is an issue of longstanding clinical and policy importance, and recent research on this association has sparked renewed debate. The author formulates six statements on the association that seem warranted by recent investigations and reviews the research evidence. In general, contrary to findings of earlier research, an association does appear to exist between mental illness and the likelihood of being involved in violent incidents. A dual diagnosis of mental illness and substance abuse probably significantly increases the risk for violence, and the association between mental illness and violence is probably significant even when demographic characteristics are taken into account. Given the considerable limitations of current research, priorities for future research include attention to the strength of the association for individual subjects, inclusion of adequate comparison groups of non-mentally-ill persons and a broad range of variables, and intensive studies of repetitively violent individuals over time.